The false-negative rate of the irrigation smear, a selfadministered vaginal cytological screening device, was determined in a group composed primarily of women in whom conventional methods of cytological diagnosis had produced findings positive for cervical neoplasia. Of the 274 examinations performed, 172 (63%) were satisfactory for cytological diagnosis. The false-negative rate in 48 cases of presumptive carcinoma in situ and invasive carcinoma was 50%. In 13 cases of histologically proved carcinoma in situ and invasive carcinoma, it was 39%. In 70 cases of presumptive cervical dysplasia, it was 51%. Possible explanations for these findings are discussed. Until the high proportion of unsatisfactory smears and the high false-negative rate of the satisfactory ones can be lowered, it is suggested that the use of this method be restricted to populations that cannot be reached by other cytological techniques.